By Sean Mahoney, Vice President, Ensono Digital & John Treadway, CEO, AI Technology Partners Hype aside, AI adoption is a journey—not a destination. These 10 best practices will help you avoid the pitfalls of tech revolutions and get on the path to sustained growth and success.
Accompanying this frenzy is an overwhelming sense of urgency reminiscent of the cloud revolution of the early 2000s. Corporate leaders are feeling this pressure from all sides (see “Fueling the fire. Feeling the heat,”). And in many ways, timing is of the essence. However, mirroring past experiences, the risks of moving too fast without thorough planning can result in a multitude of costly and damaging issues.
While AI adoption is an unquestionable imperative for every company operating today, doing so successfully requires a balance of ambition, strategy, discipline and action. By implementing the following step-by-step approach (ideally built on the foundation of a unified data platform, see page 7 to learn why this is so important), your organization can ensure maximum near- and long-term business impact with appropriate speed and minimized risk.
Just as with cloud adoption, not everyone in your organization will immediately support an AI program. If left unchecked, skeptics and active resisters can hinder the progress of AI initiatives. Identifying them and addressing their concerns is crucial for the success of your AI adoption. The key lies in early engagement.
Assembling key stakeholders early can effectively inoculate your organization against the virus of skepticism. Consider holding a structured three-day workshop as the first step in your AI Adoption Program (AIAP). The workshop can be an emotional journey for participants. There will be fear, uncertainty and doubt (FUD) about the future, particularly related to job security. As a leader in AI adoption, it’s your role to confront these concerns and dispel the FUD.
The AI Adoption Workshop should involve gathering all decision-makers, influencers and stakeholders for three full days. While this is a significant time commitment, consider the substantial investment your AI journey will require—it makes sense to invest time upfront to prevent larger, more costly mistakes later. Depending on the size of your organization and ecosystem, an AI Adoption Workshop could include as many as 25 to 30 people each day, with participants coming and going based on the topic being discussed. Here is a list of roles that should be present:
Executive sponsors – CTO, CIO and CEO, and other executive team members whenever possible
AI outcome (or program) owners – Business units, development teams
Data security and privacy – CISO, SecOps people
GRC – Governance, risk and compliance experts
Finance – Procurement, risk and governance experts
Lead architects – AI and existing infrastructure leaders
Data management – Lead data managers, data architects
IT operations – Leaders, key department heads, networking specialists
Securing commitment for all or part of the AI Adoption Workshop from everyone can be challenging, but every stakeholder’s involvement is necessary for a successful AI program. Alignment is the most crucial first step in any major IT initiative, especially one as transformative as AI adoption.
Alignment is the most crucial first step in any major IT initiative, especially one as transformative as AI adoption.
AI-first means that all of your decision-making processes, applications and data strategies should consider AI unless there’s a compelling reason not to. Without an AI-first strategy, you are simply keeping your teams tethered to traditional methods while asking them to innovate, solving the business problems you have today without anticipating the ones you may face in the future. This situation often leads to mediocre results as there is no focused dedication to making the changes necessary to reap the full benefits of AI.
The foundation of an “AI-first” commitment lies in answering the question: Why are you moving towards AI? This may appear straightforward, but it’s a question that often perplexes leaders. Plan to devote a significant portion of the AI Adoption Workshop to discussing, debating and arguing the merits and benefits of AI adoption. Without understanding why you are moving towards AI, an AI-first strategy can quickly unravel. Team members may pursue different and conflicting paths, unaware of how their actions impact others.
On the surface, AI-first might seem like a bold stance. However, without an AI-first strategy, you won’t be able to allocate the necessary resources to fully drive the organizational change needed to make a significant difference. Consider all the attendees needed for an AI Adoption Workshop. AI adoption will affect nearly every aspect of your organization. Therefore, it’s more of a strategic direction and leadership initiative than a purely technological decision.
Likewise, AI-first requires a real commitment across the organization. Any initiative without funding and committed programs and resources will struggle to achieve meaningful outcomes; AI initiatives are no different. This commitment needs to come from the top, with executive sponsorship, allocated budgets and board-level buy-in on goals, timelines and KPIs. Without significant investment, visibility and leadership focus, progress is likely to be slow and uneven and employees may not take it seriously.
When an organization truly understands the benefits of an AI-first strategy, proposed initiatives become more compelling and easier to fund.
The successful implementation of AI within your organization hinges on a solid business case. This business case must articulate the value proposition of AI in clear, quantifiable terms. It should identify KPIs and show how AI can enhance those metrics. To help identify the potential value of AI for your organization, ask questions like:
How can AI improve our products or services?
Where can AI streamline operations?
How can AI generate new revenue streams?
Once the business case is established, it’s essential to manage AI economics and return on investment (ROI) continually. AI economics refers to the cost-effectiveness of your AI initiatives. It involves understanding the direct and indirect costs of implementing and maintaining AI solutions, as well as the benefits these solutions bring.
Thus, creating and maintaining a robust framework for measuring AI ROI is crucial. Step 9 outlines the process of building this framework, including the costs you need to consider.
On the benefit side, consider factors like improved efficiency, enhanced customer experiences, increased revenue and reduced costs in other areas of the business, and determine how you define success in these areas, too. Put numbers against it, write narratives that help those around you understand what a “point of arrival” will feel like to all and, by any means necessary, quantify what success looks like.
You’ll find that costs and benefits often shift from original expectations, sometimes in ways that were unplanned. Review these shifts with executive leadership and ensure that they still represent a worthwhile endeavor. When things change, understand why and begin to steer results in the proper way, leveraging your framework for AI ROI.
The adoption of AI will significantly impact your organization, revolutionizing processes that may have remained static for years. For the first time, decision-making processes and operational workflows can be driven by AI, which comes with a mix of exciting possibilities and daunting challenges.
The adoption of AI across the organization has many implications to how business functions operate and govern their activities. Given that areas such as sales, marketing, product and operations are deeply interrelated, this underscores the need for a centralized clearinghouse and coordination function, such an AI steering committee or a full-blown Center of Excellence (CoE). Let’s explore this from the AI CoE perspective, though other models may be more appropriate for some organizations.
The AI CoE serves as the central hub for decision-making and communication for your AI program, both internally and externally. It goes beyond being a mere “AI center of expertise”; the AI CoE is a permanent operational and governing body that guides all aspects of your AI program, from the first implementation to ongoing operations.
Members of the AI CoE fall into two categories: full-time and part-time. Full-time AI CoE members are leaders who have daily responsibilities for the successful adoption, implementation and management of AI in your organization.
These include: AI program leadership, technical operations leadership, Chief Data Scientist(s), and data privacy and security leadership.
Part-time AI CoE members are leaders who have a vested interest in the success of the AI program and need visibility into the process. These include:
Legal and risk leaders
HR leaders, procurement
IT Finance
Board of Directors representative(s)
AI project owners and business units (business units may have a full-time role for a short duration during their onboarding process)
The agile nature of AI technology and its near-universal applicability completely alter how organizations operate and make decisions. Furthermore, AI-driven environments demand a tighter, more cohesive team to break down silos.
As you are integrating operations, development, data management, risk and finance, you need a central set of processes. These include: project management, technical decisions, project owner onboarding, AI and data science training, risk/security decisions, organizational change management and training, financial governance, operational services and governance, and vendor management.
AI adoption will affect nearly every aspect of your organization— it’s a strategic business priority, not a technology decision.
Identifying the right use cases is critical for AI adoption. It’s essential to analyze your organization’s processes and systems to find areas where AI can bring the most value. This involves understanding your business operations, workflows and existing systems in depth. Drill deeper into the areas of opportunity you identified in your business case to see where AI could enhance efficiency, effectiveness or customer experience. Consider tasks that are time-consuming, prone to human error or require sifting through large amounts of data. These are typically areas where AI can add substantial value.
Once you’ve identified potential AI use cases, prioritize them based on factors such as:
Business impact – How significantly will the use case affect key business metrics?
Feasibility – Do you have the data, resources and technical capabilities needed to implement the use case?
ROI – What’s the expected return on investment for the use case?
After prioritizing, select a few use cases to pilot. Pilots allow you to test your AI solutions on a small scale before rolling them out more broadly. They also provide an opportunity to learn and adjust your approach based on real-world experience. In addition to identifying use cases for AI adoption, consider how AI could help you re-envision existing processes and systems. AI isn’t just about automating existing tasks; it can also enable entirely new ways of doing business.
As you progress through this process, document your findings, decisions and outcomes. This documentation will provide a valuable resource for learning and improvement. It will also provide a record of your AI journey, helping to demonstrate the value of AI to stakeholders. This process should be methodical and iterative—it’s about continuous learning and improvement. As you gain experience with AI, you’ll likely discover new use cases, and your ability to implement AI solutions will improve.
Consider tasks that are time-consuming, prone to human error or require sifting through large amounts of data. These are typically areas where AI can add substantial value.
After identifying and prioritizing AI use cases, the next step is to implement AI solutions. We recommend using a Minimum Viable Product (MVP) approach for this implementation. An MVP is a version of a new product that allows a team to collect the maximum amount of validated learning about customers with the least effort. In the context of AI, an MVP might be a simple AI model that addresses a specific use case.
The MVP approach allows you to quickly test your AI solution in the real world. It provides valuable feedback and learnings that can inform further development. Here are the key steps in the MVP process:
Build – Develop an AI solution that addresses your chosen use case. The solution should be as simple as possible while still solving the problem at hand.
Measure – Deploy your AI MVP and monitor its performance. Use the KPIs identified in your business case to measure success. Collect feedback from users.
Learn – Analyze the data you’ve collected. What worked well? What didn’t? How can you improve your AI solution?
Iterate – Use your learnings to improve your AI solution. Then go back to step 1 and repeat the process.
The MVP approach allows you to learn quickly, reduce risk and avoid wasting resources on AI solutions that don’t deliver value. It also helps build momentum and demonstrate success early in your AI journey, which can help secure ongoing stakeholder support. Embrace the MVP mindset and be ready to iterate and adapt as you go.
The quick wins and accompanying benefits this AI Blueprint can deliver are essential for building momentum, gaining stakeholder buy-in, and paving the way for larger strategic AI initiatives.
But for AI integration to truly succeed, a comprehensive data platform and strategy is crucial. If your organization isn’t sitting on such a foundation, the time to start building it is now.
Data is the lynchpin of successful AI capability. A solid data platform will enable the most valuable AI use cases while reducing the costs and risks of all AI initiatives. Without this platform, you won’t be able to tap AI's full potential, and long-term scalability and integration across the enterprise will remain a challenge.
Balancing quick AI wins with the development of a robust data platform is vital for any organization aiming to harness AI's full power and stay competitive in the digital landscape.
AI introduces a new set of risks that must be managed, ranging from data privacy and security concerns to ethical considerations. An AI security & governance gap assessment involves identifying potential security and governance risks associated with your AI initiatives and evaluating your organization’s readiness to manage these risks. Here are some of the risks you should consider:
Data privacy and security – AI models often require access to sensitive data. How will you ensure this data is used and stored securely? How will you comply with data privacy regulations?
Model transparency and explainability –AI models can be complex and difficult to understand. How will you ensure transparency and explainability, particularly for AI models used in decision-making?
Bias and fairness – AI models can inadvertently perpetuate or exacerbate biases present in the training data. How will you ensure your AI models are fair and unbiased?
Reliability and robustness – AI models can behave unpredictably when faced with unusual inputs or changing conditions. How will you ensure your AI models are reliable and robust?
Regulatory compliance – Different industries have different regulatory requirements for AI. How will you ensure compliance with relevant regulations?
Once you’ve identified potential risks, prioritize them based on factors such as the potential impact of the risk, the likelihood of the risk occurring and related regulatory requirements. This will help you focus your efforts on the most critical risks.
Next, evaluate your organization’s readiness to manage these risks. Do you have the necessary policies, processes and tools in place? Do you have the necessary skills and expertise? Where are the gaps?
Finally, develop a plan to address these gaps. This might involve investing in new tools, hiring or training staff, or updating policies and processes. Be sure to include this in your overall AI budget.
By proactively assessing and addressing AI security and governance risks, you can build trust in your AI initiatives and ensure they are sustainable and compliant.
As AI becomes more integrated into business operations, it’s essential to implement a comprehensive AI compliance framework. This framework ensures that AI initiatives adhere to both internal policies and external regulations.
Here are the key steps in defining and implementing an AI compliance framework:
Understand the regulatory landscape – First, familiarize yourself with the legal and regulatory environment surrounding AI in your industry. This includes data privacy laws, industry-specific regulations and any AI-specific regulations that may apply to your organization.
Define AI policies and guidelines –Based on your understanding of the regulatory landscape, define internal policies and guidelines for AI. These should cover areas such as data usage and privacy, model transparency and explainability, bias and fairness, and reliability and robustness. (See also, “Leveraging your brand as you dive into the technological unknown,” The Maven Report, Spring 2023.)
Establish compliance controls – Identify controls that will ensure compliance with your AI policies and guidelines. These might include technical controls (like access controls for data), procedural controls (like approval processes for new AI models) and auditing controls (like regular reviews of AI initiatives).
Implement compliance processes – Establish processes to implement and monitor these controls. This might involve changes to your IT systems, updates to workflows or the introduction of new roles or responsibilities.
Train your team – Make sure everyone involved in your AI initiatives understands the importance of compliance and knows how to comply with your policies and guidelines. This might involve formal training sessions, written documentation or one-on-one coaching.
Monitor and audit compliance – Regularly monitor and audit your AI initiatives to ensure ongoing compliance. This might involve internal audits or external audits by third-party auditors.
Review and improve – Compliance is not a one-time task but an ongoing process. Regularly review and update your AI compliance framework in response to changes in the regulatory environment, new insights into AI risks or lessons learned from compliance issues.
By implementing a robust AI compliance framework, you can ensure that your AI initiatives are not only effective but also compliant and trustworthy.
Implementing a robust AI cost management framework is critical to preventing costs from escalating and thereby undermining the value of your AI initiatives. This framework should allow for ongoing tracking of costs and benefits and should be flexible enough to adapt as your AI initiatives evolve. Here’s how you can do it:
Understand – The first step in managing AI costs is understanding them. AI costs can include data acquisition and preparation, computational resources, software and tools, talent acquisition and retention, training, and change management.
Forecast – Once you understand the different types of AI costs, forecast for your planned AI initiatives. This should be done as part of your business case development process.
Budget – Include your AI cost forecast in your budget. Ensure that you have adequate funding for your AI initiatives. Remember to account for both upfront and ongoing costs.
Track – Implement processes and tools to track actual AI costs against your forecast. This will help you identify cost overruns early so you can take corrective action.
Optimize – Continually look for ways to optimize your AI costs. This might involve improving your data preparation processes, optimizing your AI models to reduce computational requirements or negotiating better terms with your AI tool providers.
Allocate – If your AI costs are shared across multiple departments or business units, implement a cost allocation process. This will help ensure the costs of your AI initiatives are fairly distributed.
Review and improve – Regularly review your AI cost management practices and look for ways to improve. As your experience with AI grows, you’ll likely find new ways to manage and optimize your AI costs.
While the upfront costs of AI investment can be high, the long-term benefits—when properly managed and measured—can significantly outweigh them. By implementing a robust AI cost management framework, you can ensure your AI initiatives deliver value while staying within budget.
After successfully implementing AI solutions for prioritized use cases, monitoring their performance and validating their effectiveness, it’s time to scale these solutions across the organization. Scaling involves extending the use of AI solutions to additional processes, departments or business units. It also involves refining and expanding successful AI models based on what you’ve learned from your initial implementations and pilots. Here are the key steps in the scaling process:
Review and learn – Before you begin scaling, review your AI initiatives to date. What worked well? What didn’t? Use these insights to refine your approach.
Plan your scale-up – Identify where and how you will scale your AI solutions. This might involve extending a successful AI model to new areas or developing new models based on the same underlying technology.
Prepare your organization – Scaling AI often involves significant organizational change. Prepare your organization for this change, which might involve training, change management initiatives or adjustments to roles and responsibilities.
Implement and monitor – Implement your scale-up plan and monitor the results. As always, be prepared to learn and adjust as you go.
Rinse and repeat – Even after you’ve scaled your AI solutions, continue to look for new opportunities to leverage AI. Keep up with advancements in AI technology and best practices, and continually iterate on and improve your AI models.
Implementing AI is essential in today’s business environment, but the goal is much bigger than that. By taking a systematic, iterative approach to AI adoption, you can continually learn, improve and create a culture that leverages AI to deliver ongoing value.
Act to serve today’s needs, with tomorrow’s outcomes in mind.
Yes, our organization has a well-defined strategy and clear communication regarding AI adoption.
Somewhat, we have made efforts, but there's room for improvement in addressing skepticism.
No, our organization faces significant challenges in addressing skepticism and resistance to AI adoption.
I'm not sure/I don't have enough information about our organization's efforts in this area.
